Comparison of REMS, NEWS, qSOFA and SIRS criteria scales for sepsis prediction in patients with diagnosis “SARS-CoV-2, virus unidentified”: a retrospective observational study

Background. Despite ample research on the coronavirus infection sequence and therapy, the incidence of adverse outcomes remains very high. Sepsis stands among the major factors greatly complicating treatment and increasing the risk of death. A timely identification of highrisk sepsis patients is a cornerstone of effective sepsis prevention.
Objectives. A comparative prognostic power assessment between the quick Sequential Organ Failure Assessment (qSOFA) scale, National Early Warning Score (NEWS), Initial Prehospital Rapid Emergency Medicine Score (REMS) and the Systemic Inflammatory Response Syndrome (SIRS) criteria for sepsis detection in anaesthetic intensive care patients with a diagnosis: SARS-CoV-2, virus unidentified.
Methods. A retrospective observational study included 166 patients over 18-year age with unconfirmed infection (ICD-10 code U07.2). The qSOFA, NEWS, REMS and SIRS point estimates were obtained from each patient. The patients were retrospectively divided in two cohorts by sepsis presence (Sepsis-3 criteria) to determine the express scales power in evaluating the risk of sepsis (estimated as area under ROC curve, AUROC).
Results. Data on 102 patients were included in the final analysis. Fifty-eight (57%) patients were terminal, and 55 (54%) developed sepsis. The estimates are as follows: NEWS — AUROC 0.848 [95% confidence interval (CI) 0.764–0.912], sensitivity 76.36% [95% CI 63.0–86.8], specificity 82.98% [95% CI 69.2–92.4], optimal cut-off threshold >5 points; qSOFA — AUROC 0.700 [95% CI 0.602–0.787], sensitivity 76.36% [95% CI 63.0–86.8], specificity 61.70% [95% CI 46.4–75.5], optimal cut-off threshold >0 points; REMS — AUROC 0.739 [95% CI 0.643–0.821], sensitivity 69.09% [95% CI 55.2–80.9], specificity 65.96% [95% CI 50.7–79.1], optimal cut-off threshold >5 points; SIRS criteria — AUROC 0.723 [95% CI 0.626–0.807], sensitivity 98.18% [95% CI 90.3–100.0], specificity 31.91% [95% CI 19.1–47.1], optimal cut-off threshold >0 points.
Conclusion. The NEWS scale revealed a good prognostic power to estimate the risk of sepsis in patients with suspected COVID-19 disease. The qSOFA, REMS scales and SIRS criteria possess a good calibration capacity, albeit insufficient resolution, which limits their prognostic value